When a Tax Story Gets Labelled Tennis: Lessons on Data Garbage in Sport

A news report about income-tax refunds in Pakistan, labelled "tennis", and a tennis analysis pipeline nearly swallowed it whole. I read that document on a late-weekend morning, second coffee still hot, and it took me three minutes to believe my own eyes: no player, no scoreline, no draw sheet — only the Federal Board of Revenue, the Karachi Tax Bar Association, and a filing portal called IRIS.

In thirty-seven years of reporting sport, I have seen many kinds of error. I have watched a reporter get a player's name wrong in a midnight wire copy. I have watched a sub-editor paste one match's headline onto another match's body. I have watched a major agency publish last season's team sheet on its homepage for forty minutes before correcting it. But I had never seen an automated system, designed to analyse tennis tactics at expert level, accept a fiscal-policy document and begin reasoning as though it were a Grand Slam final.

That is worth talking about, and it has nothing to do with tennis.

The document I was handed — let me describe exactly what happened — was divided into nine analytical dimensions. The first was technical and tactical analysis. The second was data and form. The third was tournament system and schedule. The fourth was tour landscape and player positioning. The fifth was rules and governance. The sixth was team and player management. The seventh was risk. The eighth was media narrative and expectation. The ninth was tennis industry transmission. Each dimension had tables, metrics, and an assessment column.

And every cell in all nine dimensions said three words: not applicable.

The analyst who did the work made a decision I respect enormously: they refused to invent. They did not construct a star player, did not manufacture a scoreline, did not write a paragraph about anyone's serve. They pointed out that eighteen of the twenty-eight information points in the source concerned refund procedures, section 7E refund applications, Sales Tax General Order No. 25/2026, and the requirement to issue a prescribed stock-transfer note when moving goods between premises. No player. No tournament. No ranking points.

This is the crux the Vietnamese sports industry should read carefully: the quality of an analytical system is decided by its capacity to refuse, not its capacity to produce.

I once wrote about a German U21 side at a European tournament in 2026 and charted every movement of its central midfielders across fourteen matches in two seasons. I sat down after each game and logged recovery positions, counted regains in the opposition third. The average came out at 11.4 per match, forty percent above the tournament baseline. I predicted high pressing would become the standard. But suppose, that day, someone had handed me a tax article and called it football, and I had still sat down, still opened the spreadsheet, still started counting — what would my published piece have been?

It would have been very fluent. It would have had numbers. It would have had charts. It would have had tactical analysis. And it would have been entirely meaningless.

That is the most frightening thing in this story. The document I read was not wrong in structure. It was better organised than many sports articles I have read. It layered clearly, it flagged confidence after every inference, it labelled risk levels. Anyone skimming the opening page would take it for a professional match report.

But behind that professionalism sat an intake system that had failed at the very first step: domain classification.

I call it the cheapest step and the easiest to skip. In a newsroom it is the act of filing a story under the right section. The person doing it is often the least experienced in the room, or today, a classification model running on a server. When it works, nobody praises it. When it fails, the consequences can propagate through the entire production line behind it.

And the consequence here was specific: two days before any human read this article, it already existed as a tennis product. Not a tax story mislabelled. A tennis product generated from a source containing not one gram of tennis material.

Which brings me to a more serious question, and I think it is the one Vietnamese sports people need to answer before building further automation.

We have a tendency to judge an analytical system by the volume of content it produces. A system generating thirty pieces a day beats one generating five. A system analysing nine dimensions beats one analysing three. But if that system has no refusal mechanism, then mathematically it is not an analytical tool — it is an amplifier, and it will amplify everything that should have been stopped at the door.

The analyst in that document did exactly one thing right: they said they could not perform the task, and explained why. That is the written form of deliberately hitting the ball out — one point lost, whole structure preserved.

I recognise that I learned this in my hardest days. In 2026, after the final in Russia, I received seventy-eight complaints that my analysis was too cold. I answered by adding emotion at the top of every piece, adding a human detail before going into the numbers. But there was one thing I did not change, and it is the thing that saved me: I never described a match I had not watched. Never. Not with a deadline closing, not with a length requirement, not with the spreadsheet already open.

That is the boundary. Not a boundary of tone, not of angle, not of length. A boundary of whether I had actually seen the thing I was describing.

An article about tax labelled tennis is an article never seen. And if someone sits down and writes nine dimensions of analysis about it, that person is not analysing. That person is typing.

Sport is the common language of humanity, but a common language is only worth something when it speaks of what is actually on the field.

There is one small detail in the document I kept, because it gave me a little optimism. The analyst assigned the highest risk rating not to any sporting factor — there were none to rate — but to the process itself: the prospect of a wrong label propagating down into finished products. They proposed adding a gate immediately before the second stage, requiring at least one player or tournament name to appear if the label reads tennis.

It is a small gate. But it is in the right place.

I wonder how many such gates are missing from our processes today — not only in sport, but wherever a system automatically generates content that nobody checks for roots. A ranking generated from bad data. A transfer story assembled from names that do not exist. A tactical breakdown built on a lineup that never took the field.

Those will be fluent. They will have numbers. They will have charts.

And they will be entirely meaningless.

What I want to keep from this strange episode is not the incident but the response to it. One person chose not to write. And that choice, inside a system that measures only output, is the most disciplined act of resistance I have seen in years.

If Vietnamese sport is entering an era of automated storytelling, the most important skill is not teaching the machine to write better, not adding numbers, not building more analytical dimensions. The most important skill is teaching the machine when to stay silent. And that, so far, can only be taught one way: a human has to stand in front, and take responsibility for the voice.